BUTTERY, GUADAGNI. LING
ditions of t h e chick feeding study, choline chloride a t all three levels was significantly effective in both increasing growth and preventing perosis on the semipurified basal diet and in decreasing perosis a t the 400 mg/kg level in the practical diet. Although stachydrine did not effect a growth response, it did produce a significant decrease in the incidence of perosis a t the 1158 mg/kg level in both diets. Betaine and methionine did not prevent perosis to a statistically significant degree. In studying the various responses of methyl donors in animals. Moyer and d u Vigneaud, (1942) fed to rats a n amino acid diet devoid of methionine and cystine but containing homocystine and vitamin supplements. On this diet, choline chloride increased growth. Betaine, when added to a similar diet (Chandler and d u Vigneaud, 1940; d u Vigneaud e t al., 1939), increased growth, but choline was more effective t h a n betaine. Jukes and Stokstad (1952) obtained similar results on low vitamin BIZ diets with and without vitamin BIZadditions. In looking a t the problem of perosis, Jukes (1940) found t h a t choline, but not betaine, prevented perosis in turkeys. T h e diet fed was not deficient in methionine or cystine. In studying compounds acting as methyl donors in man, Ciusa and Nebbia (1948) identified stachydrine, choline, betaine, and several other compounds as effective methyl donors. The diets used in this present study meet KRC requirements of methionine and do not show great differences in growth response among the supplements. In fact, in the practical type basal diet. the average weight in each supplemented group did not differ statistically from the control. However, choline and stachydrine both effected a decrease in the incidence of perosis. Jukes (1971) has discussed the question of why betaine and methionine are ineffective as substitutes for choline in preventing perosis in chicks and turkeys. He suggests that choline may function
directly in bone formation as opposed to functioning as a precursor of betaine and methionine in the prevention of perosis. The results of our study indicate t h a t stachydrine decreases the incidence of perosis, but no mechanism can be postulated a t this time.
.
LITERATL'RE CITED American Dehydrators Association, "A Study of the Major Sutritional Constituents of Dehydrated Alfalfa," Kansas City, Missouri. 1965. Beattie, F. J. R.. Blochem J 30, 1554 (1936). Bickoff. H. IM.. Soencer. R. R.. Witt. S. C.. Knuckles. B. E.. Starke. J . B . , J . A'gr. Food. Chem. 16,256 (1968). Binger. H. P.. Thompson, C. R.. Kohler, G . O ; , 1.. S. D e p . Agr. rech. Buii. no. 1235 (1961). Bregoff, H. M., Roberts, E.. Delwiche. C. C., J . Bioi. Chem. 205, 565 ( 1953). Chandler, J . P., du Vigneaud. V . . J . Bioi. Chem. 135, 223 (1940). Christianson, D. D., Wall. J. S., Dimler, R. J., Senti, R. R., Anal. C'hem. 32,874 (1960). Ciusa. W.. Nebbia. G.. Bul. Inst. Politeh. Iasi 3, 704 (1948). Duncan, D. B.. Biometrics 11, .1 (1955). du Vigneaud, V., Chandler, J. P., Moyer, A. W..Koppel, D. M., J . Bioi. ('hem. 131,s; (1939). Focht. R. L.. Schmidt. F. M.. Dowling, B. B.. J . A p r . Food Chem. 4,546 (1956). Jukes, T. H., Fed. Proc. 30, 155 (1971). Jukes, T. H., J . Nutr. 20,445 (1940). Jukes, T. H., Stokstad, E. L. R., J . ,Viutr. 48,209 (1952). Livineston. A. L.. Knowles. R. E.. Nelson. J . W.. Kohler. G. 0.. J X c r Food C'hem 16.84 (1968ai. Livingston. A. L., Smith, D., Carnahan. H. L., Knowles, R. E., Nelson, J. W., Kohler, G. O., J . Sei. Food Agr. 19,632 (1968b). Moores. R. G.. Greninger, D. M., Anal. C'hem. 23,327 (1951). Moyer. A . W.. du i'igneaud, V..J . Bioi. C'hem. 143.373 (1942) Stark. J . B.. Anal. Biochem. 4. 103 11962). Steenbock, ? j . , J . Bioi. C'hem '35, l(1918). Vickery. S . B., .J. Bioi. C'hem. 65,81 (1925). Wall, d . S..Christianson. D. D., Dimler. R. J.. Senti, F. R.. A n d . C'hem. 32,870 (1960). Received for review hlay 4. 1972. Accepted October 26. 1972. Reference to a company or product name does not imply approval or recommendation of the product by the U. S. Department of Agriculture to the exclusion of others that may be suitable.
Flavor Compounds: Volatilities in Vegetable Oil and Oil-Water Mixtures. Estimation of Odor Thresholds Ron G. Buttery,* Dante G . Guadagni, and Louisa C . Ling
Air to vegetable oil partition coefficients have been determined experimentally for a number of organic flavor compounds. These are shown to be, in general, of the same order as values calculated from solution-vapor theory. assuming that their activity coefficients in vegetable oil are equal to 1. A simple method of calculating the volatilities (air to mixture partition coefficients) of com-
The authors are carrying out a continuing study of the aroma and flavor of fried foods (cf. Buttery and Ling, 1972; Guadagni et a / . . 1972). An important factor in the effectiveness of various aroma compounds in foods is their volatility in the food medium. This is controlled to a considerable extent by the affinity of the compounds for the particular medium(s) in the food. Western Regional Research Laboratory, Agricultural Research Service. L. S. Department of Agriculture, Berkeley, California 94710. 198
J. Agr. Food Chem., Vol. 21, No. 2, 1973
pounds in vegetable oil-water mixtures is derived and shown to compare reasonably well with experimental results for such mixtures. A method is also developed for calculating odor thresholds for compounds in vegetable oil solutions from their known thresholds in water solutions. This is shown to give values which are of the same order as experimentally determined values.
With complex systems such as we have in foods, it is useful to have some model systems to relate to. Studies of these model systems can a t least give us a n approximation of the behavior we might expect in the actual practical system. T h e authors have previously studied the volatilities of a number of flavor compounds in a pure water medium (Buttery e t a l , 1969, 1971). This could be considered as one type of model system. A second important model system with foods could be the volatilities of various flavor compounds in vegetable oil. A third model system could be t h a t for water-vegetable oil mixtures. Some
FLAVOR COMPOUNDS AND VEGETABLE 011.
properties of t h e latter two systems are reported in t h e present paper. EXPERIMENTAL SECTION M a t e r i a l s . Organic compounds were obtained from reliable commercial sources. They were purified by gas-liquid chromatography before use. T h e vegetable oil used throughout the work was higholeic safflower oii "Oleinate 181" obtained from Pacific Vegetable Oil Corporation, Richmond, Calif. It was stored under nitrogen, in the dark, a n d a t refrigerator temperatures. G a s C h r o m a t o g r a p h y . T h e method used was essentially t h a t described previously for water solutions (Buttery et al., 1969, 1971) However, in t h e case of water solutions, it was possible t o inject some of t h e solution for comparison with the vapor peak. With oil solutions, a standard solution of the material in water, hexane, or ether was used for comparison with the vapor peak. O d o r Thresholds. Odor thresholds were determined essentially as described previously for water solutions (Guadagni e t al., 1963, 1972) using teflon bottles and t u b ing. In this case Oleinate 181 vegetable oil was used as solvent. RESULTS AND DISCIJSSION Many authors have studied some aspect of the volatility of flavor compounds in foods. This work has been previously reviewed b:y the authors (Buttery e t a / . , 1969). For our model systems we can consider a closed teflon bottle approximately half full of the liquid medium under consideration. We will consider only dilute solutions less than about 1% of single organic compounds dissolved in the medium, and a t a temperature of 25" a n d a t atmospheric pressure. We will consider also t h a t the bottle is swirled gently (horizontally) t o facilitate the attainment of a n equilibrium Our main interest is in being able to relate the concentration of a compound in the headspace to its concentration in solution. Much can be understood from established t:heory on solutions and volatility. An experimental approa.ch is also desirable, partly because of t h e lack of fundamental data, but also to verify the validity of theoretical calculations. T h e most practical factor for food aroma work is t h e air-to-solution partition coefficient. This is, of course, the ratio of solute concentration in t h e atmosphere above the solution to t h a t in solution, a t equilibrium, as indicated in eq 1. This can be related to the Henry's law constant C, as shown in eq 2. air to solution partition coefficient
In previous work we have shown that air-to-water partition coefficients calculated using these types of equations compared fairly well t o experimental values (Buttery e t al., 1971). Activity coefficients for aliphatic aldehydes in water solutions can vary from a value of 4 . 2 for acetaldehyde to 71,000 for nonanal (Bomhen and Merson, 1969). These activity coefficients were calculated l'rom literature d a t a (Pierotti e t al., 1959). Unfortunately very lew such values are available for vegetable oil solutions. Some are available for paraffin oil, but this has very different solubility properties t h a n vegetable oil. Experiments by the authors showed, however, that vegetable oil was a reniarkably good solvent for a wide variety of food aroma conipounds. In fact many such compounds seemed completely miscible with vegetable oil. In this case -Vs. the solubility in mole fraction units, is equal to 1. The activity coefficient y generally also approaches 1. We should point out, however, t h a t eq 4 is only meant for cases where the solubility is low. From our experimental results it seems that for many common flavor compounds, the activity coefficient in vegetable oil solutions is a t least of the order of 1. which is quite different from the very large variation found in water solutions. Table I lists experimental air t o vegetable oil partition coefficients as determined by methods previously described by the authors (Buttery et ai., 1971). For coniparison. the values calculated assuming that the activity coefficient y is equal to 1 are also listed. Except for pentanol, the calculated and experimental values are of' the same order. With hydrophilic compounds such as the aliphatic alcohols and probably also free organic acids. etc.. we might expect y to be considerably greater than 1. especially for the lower homologs. I n the case oi pentariol, -J appears to be roughly 18. It is interesting to compare experimental air-to-water partition coefficients with air-tovegetable oil partition coefficients for homologous aliphatic aldehydes. as is shown by the solid lines in Figure 1. The relations show quite a contrast. For a compound such as butanal, there is not very much difference between the values for oil and water. but with octanal the difference is of the order of 1000. Vegetable Oil-Water Mixtures. In many foods we have reasonable amounts of both oil (or f a t ) and water as the two main liquid phases. Milk is a good example of this. With a n equilibrium of a solute between an oil-water mixture and air. we have a three-phase system. This is more complicated t h a n a two-phase system. but still can be handled fairly easily. Using thermodynamic quantities. Bomben a n d Merson (1969) have developed equations for handling such mixtures. Following a similar line of reasoning to these authors, a somewhat simpler relation can
K,, = (solute concentration in air)/ Ka3 = C
(solute concentration in solution) (1) (solvent conversion factor) (2)
X
T h e solvent conversion factor is a simple number which involves conversion of pressure units to mass units using gas laws a n d conversion of solvent volume to mole fraction terms. This factor is dependent on the molecular weight of the solvent but is independent of the solute molecular weight which cancels out in t h e calculation of this factor. For water this factor is 0.97 x 10-6 and for vegetable oil it is 5.2 x As outlined previously (Buttery e t a l , 1971) from solution-volatility theory
c
=PoXY
C = p ( , X 1 / N > ( i f solubility is low)
(3) (4)
where p o is the vapor pressure of the pure compound. y is t h e activity coefficient in t h a t solvent, and N , is the solubility of the compound in t h a t solvent in mole fraction terms.
Table I. Air to Vegetable Oil Partition Coefficient (25") Determined Experimentally and Calculated Assuming the Activity Coefficient (y) is Equal to 1
-
Air to vegetable oil partition coefficlent
Compound
Experimental
Calculated assuming *
Butanal Pentanal Hexanal Heptanal Octanal Butan-2-one Heptan-2-one But-2-enal Hex-2-enal Pentanol 2-Methylpyrazine
2.3
X
4 7
1.0 x 1 0 - 3 3.5 x 1 0 - - 4
20
1.0 4.0
x
10-4
3
XlO
XlO X10 5 2 XlO 078x10 1 8 X10 2 4 x10 1 3 x10 5 2 x10
88
X
1.9 x 10-3 1.03 X 1 . 4 x 10--3 1.6 x 1 0 ~ - 4 2.3 x 10-3 2.0 X
J. Agr. Food
6 2 24
x10 x10
Chem., Vol. 21, No. 2.
1973
199
BUTTERY, G U A D A G N , LING
CH,(CH,),CHO
Dividing both top and bottom lines by C, gives us eq 7.
K3''' = C,, xF, ~-
1 ____ COIx F,,
(7)
+-
C, C, We can call the air-to-water partition coefficient (C,/ C,) K,, and we can call the air-to-oil partition coefficient (Ca/Ccll)KOI.Equation 7 is then simplified to
CARBON NUMBER
Comparison of air to solution partition coefficients (25") for aliphatic aldehydes in water ( + ) and vegetable oil ( 0 ) (both solid lines) and for 1 % ( 0 )and 10% ( A ) mixtures of oil in water (broken e n d dotted lines) Figure. 1 .
be obtained in the following way. T h e air-to-mixture partition coefficient is given by: air to vegetable oil-water mixture partition coefficient
I
